Generating the IOHandler.h file out of IOHandler.cmake.hpp through CMake: That way no specific definitions need to be added to the compiler flags.

Fixing a few issues regarding the compilation of GSNIO. These are painful workarounds to allow compilation and linking.

Every former plugin now adds a definition. This define can be checked in IOHandler. This is not the best solution, because turning a plugin on or off results in compilation of the whole meteoio.

The plugins are no longer separately loadable libraries, but are fully incorporated into libmeteoio. All classes and facilities that provided capabilites for the dynamic loading were deleted (DynamicLibrary). IOPlugin now holds a function pointer to an instantiation function for children of IOInterface. In IOHandler::registerPlugins() all plugins need to be registered with the key name, the class name, a pointer to an instance of the class and a function pointer to the instantiation function for the class (child of IOInterface). IOInterface lost some complexity and thus the constructor of the base class does not have to be called in derived classes anymore.

The CMakeLists.txt files had to be updated to no longer make separate dynamically loadable libraries for each plugin, but to simply add the classes the user selectes to the meteoio library. 

As a side effect the io.ini key GENERAL::PLUGINPATH is no longer necessary.

Removing some costly but unnecessary features (ie there is no need to always keep gmt_year, etc up to date when we in any case never use the gmt values!). The gain is ~5% on the speed of data preprocessing in Snowpack.

Fixing a small bug/warning in libsmet and properly initializing the members in some plugins (through initialization lists, as usual).

By the way, the last changes to the code made it slightly faster for a 1 year data_converter (less than 3%) all while being much safer (not relying on copying pointers verbatim and hoping they remain valid).

The automatic tests have shown that the code was getting slower following the latest commits. After investigations (thanks valgrind!), it has been found that the newly added copy constructor of Coords was used a lot as well as setFunctionPointers(). Therefore the function pointers have been removed and replaced by a simple list of if(coordsy=="") etc. The run time is now back to normal (and actually up to 6% faster on some tests).

Cosmetic touch up on Date.

After compiling with -Weffc++, it appeared that lots of things could be done in initialization lists. Some of it has been done for this commit (but quite a few remain).

Some extra documentation has been added for the UPS coordinate system. The coordinates test has been expanded. A missing default value has been added to the Suntrajectory.

The Universal Polar Stereographic projection coordinates have been implemented. A small bug has been discovered (and fixed!) for latitude/longitude specifications as degrees/minutes/seconds for negative coordinates (the conversion to decimal was then wrong because of an improper rounding direction).

Consider threshold solar elevation in static projection method for consistency with the non-static one

Some better documentation, a new method to project a radiation from the horizontal to a slope (as needed by Snowpack), a foolproof method for computing the splitting (it only gets one parameter, the measured radiation, and knows which potential radiation to use...). Now Snowpack can use a Sun object for computing its radiation!

New optimized pow() replacement have been implemented. The Sun object can not use them, though (error accumulation leading to a dramatic loss of accuracy). The Sun object now checks that a given julian_date is different from the already stored one (to avoid recalculating all if not necessary).

The user provided cloudiness can now be forced to the ILWR models (if not nodata). The selection of which model to use based on the available parameters has been improved and verified.

A new method for getting the parametrized ILWR has been implemented that chose between various parametrizations depending on the available data. The data_converter example has been slightly edited for clarity.

A wrong commit on doc/examples/io.ini has been reverted. A new undercatch filter has been implemented, following (Hamon, 1973). I would not recommend using it for the moment, though. 

The issue 222 was actually more complex than it seemed: the behavior of an uncaught exception is actually not specified by the standard, so there is nothing that could be doneon MeteoIO's side: instead, each application is invited to properly wrap its calls in try/catch blocks, or wrap its whole main in try/catch. The code examples and examples have been modified to show several such possibilities.

some white spaces clean up and tagging as well as proper definitions for the upcoming visibility markup

Now testing the array1d<double>. The Undercatch filter has been renamed as Undercatch_wmo so we can offer competing models (hint: Hamon 1973).

Added some documentation to GRIBIO, fixed the documentation of FilterStdDev, improved the regex for the sun test, significantly improved code coverage by adding lots of different filters and interpolations (both 1D and 2D). The meteo files have been tweaked in order to contain precipitation for the reference time step.

- Befor each test run make a distclean
- Corrected Valgrind starting script so that CMake cache is correct initialised

Change start script, so that independent where ctest is installed, ctest runs ! (but need to be somwhere in the PATH definited locations !!)

Improved error message when reaching outside of an array in debug mode and fixed a stupid allocation bug (swapping two lines in the last commit triggered this bug)

Add the testing scripts for the cron job. A single cron job is now necessary, calling startScript.sh.